Yakpo, Kofi & Adrienne Bruyn. 2015. Transatlantic patterns: The relexification of locative constructions in Sranan. In Pieter Muysken & Norval Smith (eds.), Surviving the Middle Passage: The West Africa-Surinam Sprachbund, 135–75. (Trends in Linguistics, Studies and Monographs (TiLSM) 275). Berlin: De Gruyter Mouton.
AUTHOR MANUSCRIPT VERSION
Trans-Atlantic patterns: the relexification of locative constructions in Sranan Kofi Yakpo and Adrienne Bruyn
1. Introduction
Sranan and the other creoles of Suriname have long been noted for their use of postpositions in the expression of spatial relations (cf. e.g. Muysken 1987). This characteristic sets these languages apart from the vast majority of Afro-Caribbean English-lexifier Creoles, both in the Americas as well as in West Africa. The use of postpositions, some of which are derived from English words for body-parts, is one of the more conspicuous features pointing towards substrate influence in Sranan. Beyond this particularly visible African presence, the grammar of spatial relations in Sranan contains many more features that suggest a diffusion from Africa, and to be more precise, from the Gbe languages, as well as Western Bantu via Kikongo (for the dominant role of the Gbe substrate of Sranan, as well as the secondary role of the Kikongo substrate cf. e.g. Arends 1995; Arends, Kouwenberg, and Smith 1995; Huttar 1981, 1986; Migge 1998a, 1998b, 2000, 2003; Smith 1987; Winford 2000). The affinities of Sranan with these African language(s)/families can be traced in the semantics of individual locative elements. For example, the word baka, derived from Engl. ‘back’ is the regular form employed for the expression of the body part as well as the spatial concept ‘behind’ in Sranan. The semantics of baka overlaps with that of the Gbe (Ewe) item mègbé ‘back’ which is also employed with both senses. There is good reason to assume that such systematic correspondences in meaning and function represent cases of local relexification, that is, of individual forms. The main purpose of this chapter is, however, to show that the participation of relexified Sranan forms like baka in multi-constituent locative constructions constitute cases of pattern relexification. We will show that the concept of pattern relexification can explain the behaviour of Sranan locative elements in instances where an account based on local relexification alone would be stretched to its limits. Pattern relexification makes allowance for differences between Sranan and the substrate languages in the behaviour of individual items. The reason is that the 136 Kofi Yakpo and Adrienne Bruyn
Relexification of patterns involves the transfer of lexical properties of individual forms plus their relational properties. A central part of the argument for pattern relexification is that Niger-Congo substrate patterns manifest a large degree of homogeneity, and that this probably facilitated the relexification in Sranan of morphosyntactic blue-prints or skeletons. At the same time, we will see that Sranan locative constructions also reveal the intricate interplay of substrate patterns, patterns inherited from the lexifier English, influence from Dutch, which has served as a superstrate for more than three hundred years, as well as internal development. In this context, we should clarify our use of the terms “lexifier” and “superstrate”. Suriname consitutes a case, in which the lexifier (the language that provides the bulk of the lexicon, and that of the basic lexicon in particular) and the superstrate (the language that has serves as the language of the socially dominant group) of the creoles are not identical (cf. Selbach 2008). The ancestor language of Sranan and the other Surinamese creole languages was formed during a relatively brief period of English colonial rule (cf. Smith, this volume on the early history of Suriname), with English serving as the lexifier, and by default, also as the superstrate language. When the Dutch took control of Suriname in 1667, Dutch replaced English as the colonial language and has thenceforth also served as the superstrate (i.e. socially dominant) language of Suriname. One conclusion drawn from the data presented in this chapter is that the presence of substrate patterns in locative constructions is significant, both in a diachronic and a synchronic perspective. The strong parallels in the grammar of spatial relations between Sranan and Gbe in particular provide further support for the existence of a Transatlantic Sprachbund that unites the Surinamese creoles and the Gbe languages with respect to a substantial number of isoglosses. The Sranan examples in this chapter for which references are not provided stem from a corpus of primary data collected in Suriname and the Netherlands in 2011 by Kofi Yakpo as part of the “Traces of Contact” project of Radboud University Nijmegen. Unless indicated otherwise, examples from the Gbe languages are also from field data, collected by Kofi Yakpo in Ghana and Togo between 2003 and 2011 and speaker intuitions of Kofi Yakpo. Tone-marking is provided for the authors’ primary data and wherever contained in the sources. After providing an overview of locative elements in Sranan in section 2, we describe the expression of three important spatial relations in Sranan and Gbe in section 3. In section 4, we attempt to explain the variation encountered in Sranan locative constructions by additionally drawing on Kikongo data. Section 5 summarizes and systematizes the findings, and section 6 concludes the chapter. The relexification of locative constructions in Sranan 137
2. Locative elements in Sranan
This section provides a brief overview of the forms and functions of locative elements in Sranan. The inventory of locative elements (i.e. prepositions and locative nouns) in modern Sranan largely consists of items of English origin, with a minority of Dutch origin. However, these locative elements may appear in constructions that represent substantial departures from the corresponding ones in the English and Dutch. We conclude that the differences between Sranan on the one hand, and the English and the Dutch on the other, are largely due to substrate transfer. We base our analyses on examples from the Gbe languages Ewe (Ghana, Togo), Gen (Togo), Aja (Togo, Benin), Gun (Benin) and Fon (Benin). We should mention here that the historical evidence adduced by the authors mentioned further above suggests that Fon varieties (hence the eastern reaches of the Gbe continuum) constituted the single most important substrates of Sranan, rather than more western varieties like Gen and Ewe. However, we have found it useful to consider corresponding structures from varieties other than Fon because it shows that the templates for expressing spatial relations in all the Gbe languages are virtually identical. This strengthens the argument for a general Gbe origin of the patterns employed to express spatial relations in Sranan, since there is no need to show an exclusive, or even predominant influence of Fon. The Sranan locative elements employed for expressing the basic spatial relations relevant for the discussion are given in 6.1:
Table 1. Sranan locative elements Locative element Meaning Source language(s) ini inner part, in ‘in’ (Du./Eng.)
na doro outside ‘LOC door’ (Eng.) tapu top, on ‘(on) top (of)’ ondro bottom, under ‘onder/under’ (Du./Eng.) fesi face; in front ‘face’ baka back, behind ‘(at the) back (of)’
fu general location; SOURCE- ‘for’ (Eng.) oriented na general location ‘na’ (Igbo/Port, cf e.g. Parkvall 2001: 108)
Some of the forms in the Table have corresponding near-homophones in English and Dutch (i.e. Sranan ondro, English under, Dutch onder). Although simultaneous influences from Dutch and English, hence convergence, are in principle possible, the entire system of specific prepositions is derived from the lexifier English. 138 Kofi Yakpo and Adrienne Bruyn
We therefore assume English forms to be the source forms unless the contrary can be proven. The forms ini ‘inside’ and ondro ‘under’) are are derived from the corresponding prepositions in English. The elements baka ‘back, behind’ and tapu ‘top, on’ are only found in complex locative structures in English (e.g. at the back of the car) and have nominal uses (e.g. my back) as well. The element fesi is only found with a locative sense in specialized contexts in English but not with a general meaning as in Sranan (e.g. ‘the face of the building’). The element na doro, literally ‘at the door’ and with the meaning ‘outside’ is a Sranan neologism that has no exact correspondence in English or Dutch. Among its spatial senses, the element fu functions as a general locative preposition to denote a PLACE, however less prominently so than na, which follows below. The preposition fu may also be employed to denote a SOURCE and if this is the case, appear without support from other PATH-denoting locative elements, as in (1). One possible explanation for the PLACE and SOURCE senses of this preposition is that the English prepositions for and from may have converged into fu during the formative period of the language:
(1) ala den wroko disi wi leri fu mi papa. all DEF.PL work this 1PL learn ABL 1SG father ‘All these (types of) works we learnt from my father.’ (Sranan; Hart 1996: 17)
At the bottom of the table we find the only element without a Dutch or English etymology, namely the general locative preposition na ‘LOC’ (with its modern variant a). Reflexes of this form are present throughout the family of Afro-Caribbean English-lexifier Creoles, even if na is not found in all languages. In Sranan, the preposition na ‘LOC’ functions as a general GROUND marker and may introduce participants with PLACE (2), GOAL (3), SOURCE (4) and PATH (5) roles. In the following sections we will see how corresponding forms fulfil very similar functions in the substrate languages of Sranan:
(2) mi e tan na boiti. 1SG IPFV stay LOC countryside ‘I live in the countryside.’ [Sranan; Hart 1996: 38] The relexification of locative constructions in Sranan 139
(3) a waka esesi go na oso. 3SG walk quickly go LOC house ‘She walked to the house quickly.’ [Sranan]
(4) mamanten a komopo na oso. morning 3SG come.out LOC house ‘In the morning, he left the house.’ [Sranan]
(5) mi boro na a busi kon na oso. 1SG pierce LOC DEF.SG forest come LOC house ‘I (took a short-)cut through the forest to the house.’ [Sranan]
Na ‘LOC’ is one of the very few items in Sranan that functions unambiguously as a locative preposition (rather than alternatively, as a postposition) in a way resembling prepositions in English and other Indo- European languages. However, we will see that beyond a superficial linear equivalence of locative constructions like (3) and (4) above with English prepositional phrases like at/in/to the house, the functions of na ‘LOC’ are very different from that of any English locative preposition. Aside from that, only the two elements ini and ondro are derived from forms serving exclusively locative functions in the source languages as well. The remaining forms tapu, baka and fesi are derived from landmark and body part expressions that are not used as basic locative elements in the source language in the same way as in Sranan. What characterizes all the European-derived forms in Table 6.1 is that they may appear in morphosyntactic structures very different from English ones in the corresponding contexts. In the following sentence, the Sranan locative element ondro ‘under(part)’ co-occurs with an additional locative element, the general locative preposition na ‘LOC’. Unlike its English counterpart under, Sranan ondro may also appear in a post- rather than a prenominal position:
(6) a buku de na a tafra ondro. DEF.SG book be.at LOC DEF.SG table bottom ‘The book is under the table’ [Sranan]
We are thus confronted with a situation in which the system employed for the expression of basic spatial relations in Sranan is characterized by a substantial departure from the corresponding English and Dutch ones. In this system, Sranan locative prepositions are (a) either not derived from English locative prepositions at all (i.e. na); or (b) are derived from simple or complex English locative prepositions, are employed with locative meanings in 140 Kofi Yakpo and Adrienne Bruyn
Sranan as well but may appear in wholly different morphosyntactic structures (i.e. ini, ondro and tapu in postnominal position); (c) are derived from English body part expressions not normally employed as basic locative elements in English and also appear in wholly different morphosyntactic structures (i.e. baka and fesi in postnominal position). In the following sections, we will show that comparable strategies involving morphosyntactically and semantically similar forms are also employed for the expression of spatial relations in the substrate languages of Sranan. We also address cases in Sranan in which English-derived locative elements are used in genuinely prepositional functions. These uses are the consequence of language contact with Dutch and are not attested in earlier stages of the language. Such a development can be seen as forming part of a larger restructuring process, in which many of the typologically (West) African features of Sranan have entered into competition with Germanic features via contact with Dutch (cf. eg. Essegbey and Bruyn 2002; Essegbey 2005).
3. Locative constructions in Sranan and Gbe
In this section, we will be concerned with two types of spatial descriptions: static location involving a GROUND with a PLACE role and motion events involving GROUNDs with a GOAL and a SOURCE role. We suggest that the overall picture with respect to locative constructions in Sranan and the substrates is one of unity in diversity. This means that we often find non-negligible local differences between Sranan and substrates in the semantics and the morphosyntactic behaviour of individual elements. At the same time, striking similarities in the semantic organization and the morphosyntactic realizations of the relevant spatial relations can be observed at a higher, paradigmatic and syntagmatic level. We conclude that this similarity in patterns is due to relexification. Henceforth, the term “locative construction” is employed for the various structures covered here that instantiate spatial relations – whether they involve static location or motion. The term “locative element” refers to prepositions, postpositions and locative nouns alike. The following terms employed for the constituents of locative constructions are presented by means of the Sranan and Fon spatial descriptions in (7) and (8) and their English equivalent in (9) (the numbers in squared brackets refer to the numbers in superscript ): FIGURE [1] = entity located or moving; GROUND [2] = the entity which acts as a spatial reference point for the location or motion of the FIGURE; PATH [3] = the path of motion of the FIGURE to
The relexification of locative constructions in Sranan 141
(GOAL) or from (SOURCE) the GROUND; REGION [4] = the space anchored to the GROUND; (SPATIAL) RELATION [5] = relationship between the FIGURE and the GROUND, mediated through location-denoting predicates, adpositions and locative nouns (cf. e.g. Talmy 1985, 2000; Levinson 1992):
(7) mi teki moni1 komoto3 na5 a dosu2 ini45 1SG take money come.out LOC DEF.SG box inside. [Sranan]
1 3 2 45 (8) ǹ sɔ́ àkwɛ́ sìn gbàví ɔ̀ mɛ̀ 1SG take money come.from box DEF inside. [Fon; Höftmann 1993: 140]
(9) I took money1 from345 the box2
A comparison of the examples above reveals a cline in the semantic transparency of the spatial description in the three languages. Sranan manifests the highest degree of isomorphism in that most participating elements denote only one particular aspect of the spatial description. English is characterized by maximal opacity. English utilizes a single form, the preposition from conflates PATH, REGION and SPATIAL RELATION aspects. Fon is situated in the middle, with the element mɛ̀ conflating two aspects of the spatial description. The portmanteau nature of the English preposition from contrasts with the Sranan and Gbe constructions in two ways. Firstly, the PATH component of the spatial description is indicated by a PATH-denoting V2 in an SVC in the latter two languages. Secondly, Sranan and Gbe feature an additional locative element, where English only has one, namely a REGION-denoting locative element. These aspects represent two major typological differences in the realization of spatial descriptions between Sranan and Gbe on the one hand, and Germanic (and Standard Average European as a whole) on the other (cf. Creissels 2006; Heine, Claudi, and Hünnemeyer 1991: 140-143). In the remainder of this paper, we employ the term “PLACE” when referring to a spatial relation that involves a static, at-rest relation (also sometimes referred to as “essive” in the literature). The terms “GOAL” and “SOURCE” refer to the two basic motion-oriented spatial relations, namely movement towards a GROUND (also referred to as “allative”) and movement away from a GROUND (also referred to as “ablative”). 142 Kofi Yakpo and Adrienne Bruyn
3.1 PLACE-oriented relations
The first type of construction that we address involves a FIGURE located with respect to a GROUND without any motion involved. Such PLACE relations include “basic locative constructions”, which answer where-questions (Ameka and Levinson 2007). These constructions are semantically and structurally less complex than motion descriptions and we will therefore use them in order to discuss some general characteristics of locative constructions in Sranan and Gbe. We will show that the expression of a PLACE relation in Sranan is highly similar to that found in the Gbe languages, both in terms of the semantics of the elements employed as well as with respect to morphosyntax. The differences that can nonetheless be found between Sranan and Gbe can be attributed to competing substrate patterns and contact with Dutch. In the basic locative construction, both Sranan and the Gbe languages feature a locative-existential copula which is semantically rich enough to express the SPATIAL RELATION by itself. Hence neither in Sranan nor Fon do named places and other known or expected locations require to be marked by an additional locative element unless a higher degree of specificity is desired. Although the English lexifier and Dutch superstrate may feature reduced definiteness marking in these contexts, prepositions are still necessary besides the copula (i.e. hij is op school/he is at school):
(10) mè cè lɛ́ ɖò Paraku. relative 1SG.POSS PL be.at PLACE ‘My relatives are in Paraku.’ [Fon; Höftmann 1993: 189]
Sranan also employs a separate locative-existential copula in basic locative constructions. However, a difference with Gbe is that in Sranan the GROUND is additionally marked by the general locative preposition na:
(11) a de na wasi-oso. 3SG be.at LOC wash-house ‘She is in the bathroom.’ [Sranan]
A higher degree of specificity may be obtained in these constructions through the use of a ‘nouny’ locative element denoting the REGION in Sranan (cf. (12)) and the Gbe languages (cf. (13). In both Sranan and Gbe, the locative noun may be seen to function as a possessed/modified noun The relexification of locative constructions in Sranan 143 and syntactic head to the preceding GROUND NP (cf. Aboh 2010) in an “associative construction” (Welmers 1973: 283):
(12) a buku de a tafra tapu. DEF book be.at LOC table top ‘The book is on the table.’ [Sranan]
(13) nɔ̀ cè ɖò àxì mɛ̀. mother 1SG.POSS be.at market inside ‘My mother is at [in] the market.’ [Fon; Höftmann 1993: 189]
Postpositional locative nouns already occur in historical records of Sranan in such complex locative structures, cf. (14).
(14) sinsi a komm na hosso inni. since 3SG come LOC house inside ‘since she entered the house’ [Sranan; Schumann 1783]
Our corpus however only contains a handful of postpositional structures like the ones above. The overwhelming majority of locative constructions in the corpus involve prepositional locative nouns. (cf. also Essegbey 2005: 237). Prepositional structures are also already attested in Early Sranan, as shown in the following example. It is however, impossible to assess the relative frequency of pre- and postpositional structures in Early Sranan:
(15) trueh da dotti na ondro boom. throw that soil LOC underside tree ‘Throw that soil to the bottom of the tree.’ [Sranan; Schumann 1783, cited in Essegbey and Bruyn 2002]
We can conclude that Sranan locative elements have retained a large part of the phonological shape and a considerable part of the lexical information of their English etymons. At the same time, they have undergone a morphosyntactic recategorization from preposition to locative noun (i.e. ini ‘inside’) or common noun to locative noun (e.g. fesi ‘in front of’). They thereby come to resemble their Gbe counterparts much more than their English etymons. One difference between the Gbe substrate and the Sranan, is that these locative constructions, whether pre- or postpositional, whether involving motion or not, may invariably be introduced by the general locative preposition na ‘LOC’. This circumstance sets Sranan locative 144 Kofi Yakpo and Adrienne Bruyn constructions apart from the corresponding Gbe ones and will be addressed in due course. However, the use of the general locative preposition na is far from obligatory – if it was in Early Sranan this is certainly no longer the case in contemporary Sranan. In our corpus, locative structures introduced by na (or its variant a) are equally common as ones where the locative preposition is absent. Dutch influence may be held responsible for what seems to be a rather fundamental ongoing reorganization of the locative system (cf. also Essegbey and Bruyn 2002). In the corpus data, na-less structures are attested in the description of static location as well as motion events. They are found with the entire range of English/Dutch-derived locative elements listed in Table 6.1. Compare ondro ‘under(side)’ in (16), baka ‘back(side)’ in (17), ini in (18) and fesi ‘in front of, opposite’ in (19):
(16) a buku de ondro a tafra. DEF book be.at under DEF.SG table ‘The book is under the table.’ [Sranan]
(17) den dringi biri kibrikibri baka a oso. 3PL drink beer secretly behind DEF.SG house ‘They drank beer secretely behind the house.’ [Sranan]
(18) a e sidon ini a oso. 3SG IPFV sit inside DEF.SG house ‘She is sitting in the house.’ [Sranan]
(19) a sidon fesi a oso. 3SG sit in.front.of DEF.SG house ‘She sat down in front of the house.’ [Sranan]
We assume these Sranan structures to be induced by contact with Dutch because the uses of the locative elements are indistinguishable from the uses of prepositions and Dutch. Hence they conflate REGION and SPATIAL RELATION as in the four sentences above and additionally, PATH as with ini in (20) below:
(20) a e poti a spun ini a preti. 3SG IPFV put DEF.SG spoon in DEF.SG plate ‘He is putting the the spoon into the plate.’ [Sranan]
So Sranan may make use of structures for expressing a PLACE relation that are virtually identical to the ones found in the Gbe languages. In both (groups of) languages we find postpositional locative nouns instead of The relexification of locative constructions in Sranan 145 prepositions. One difference in need of an explanation is the obligatory presence of the general locative preposition na ‘LOC’ where Gbe has no corresponding element. This question will be addressed further in due course. Besides that, Sranan also features purely prepositional uses of the same locative elements that function as postpositions in other contexts. Following Essegbey and Bruyn (2002), we assume that these prepositional uses are a fairly recent development induced by contact with Dutch. In the following section, we explore further parallels between Sranan and Gbe in the expression of motion events involving GOAL- and SOURCE-oriented locative relations.
3.2 GOAL- and SOURCE-oriented relations
Locative construction that involve motion events display strong similarities in Gbe and Sranan in their overall make-up. The major Gbe characteristic reflected in Sranan motion descriptions is that verbs or verb-derived prepositions are used in functions occupied by prepositions with no verbal etymologies in English and Dutch. In Sranan and in the Gbe languages, GOAL-oriented motion events are expressed through the interaction of verb(s) and locative elements. Some Gbe directional verbs may appear in clauses in which the GOAL is expressed as a direct argument of the verb. Hence, the GOAL of the locomotion verb yì ‘go (to)’ is not preceded by a preposition or serial verb and is therefore encoded like any transitive object in the Gbe language Gen:
(21) wò yì kɔ́̃jí à? 2SG go hospital Q ‘Did you go to the hospital?’ [Gen]
In caused-motion events involving inanimate transitive objects, the general picture in Gbe is that verbid prepositions - ɖé (in Ewe) and dò (in Fon/Aja) mark the GOAL. These verbids are derived from a verb meaning ‘reach, enter’ and are glossed as ‘ALL(ative)’ in their prepositional function (cf. Ansre 1966; Aboh, Ameka, and Essegbey 2007). Aboh, Ameka, and Essegbey present a detailed analysis of these forms, which have grammaticalized into prepositions in some Gbe varieties but are characterized by residual verbiness in others (e.g. the inland varieties of Ewe). A relevant characteristic of the prepositional uses of the form ɖé/dò, which 146 Kofi Yakpo and Adrienne Bruyn may also be seen to be indicative of its advanced degree of grammaticalization, is the fact that the allative preposition may be used to mark inanimate and animate GOALS (i.e. RECIPIENTS) alike. Compare example (22), which involves the inanimate GOAL egli ‘wall’ in Aja with example (23), which involve the animate GOAL (RECIPIENT) ‘friend’ in Ewe and Fon respectively.
(22) Kojó sɔ́ eba lɔ́ xɔ do egli nu. NAME take stick DEF hit ALL wall outer.surface ‘Kojo hit the stick against the wall.’ [Aja; Morley 2008: 95]
(23) n sɛ̀ wèmá dò xɔ́ ntɔ̀ n cè. 1SG throw book ALL friend 1SG.POSS ‘I sent a book to my friend.’ [Fon; Höftmann 1993: 111]
Gbe GOAL-oriented constructions differ from SOURCE-oriented constructions in an important aspect: GOALs cannot be marked by the general locative preposition (i.e. lè in Ewe/Aja and ɖò in Fon). Given the origins of these prepositions in the locative-existential copula (which instantiates a static concept) it is not too surprising that they may not mark syntactic GOAL objects in Gbe (which represent the endpoints of a motion). We have already seen in (3) above that Sranan GOALs are canonically marked with na ‘LOC’ in GOAL- oriented motion events where Gbe may feature unmarked GOALs, as in (21) above. At the same time, Sranan does not have a Gbe-style general allative (GOAL) preposition derived from a verb meaning ‘reach’ or ‘enter’. In fact, the use of doro ‘reach’, the lexical equivalent in Sranan to the Gbe verb dò/ɖé instead of na is judged ungrammatical by my informants, cf. (24):
(24) *mi seni a buku doro mi mati. 1SG send DEF.SG book reach 1SG friend ‘I sent the book to my friend.’
Seen from a Gbe perspective, the absence of a general allative preposition in Sranan leads to a shift of the GOAL-oriented meaning component of the motion event from verb plus (verby) preposition in Gbe to common SVCs consisting of the string V1+V2 in Sranan. Hence in caused-motion events like (22) Sranan features a variety of specialized/lexicalized SVCs in which there is a (albeit limited) variability of the V2. Hence the V2 may change according to semantic factors such as animacy of the GOAL or type of contact with the GOAL. In this vein, the equivalent of the Gbe sentence in (22) above involves the verb string fringi – naki ‘throw – hit’ in Sranan: The relexification of locative constructions in Sranan 147
(25) a fringi a tiki naki a skotu. 3SG throw DEF.SG stick hit DEF.SG wall ‘He threw the stick against the wall.’ [Sranan]
Likewise, a corresponding way of rendering the Gbe example (23), which features the allative preposition with an animate GOAL, i.e. RECIPIENT, must involve the verb-derived dative marker gi (< ‘give’) in Sranan. Compare (26) with the ungrammatical example in (24) above:
(26) mi seni a buku gi mi mati. 1SG send DEF.SG book DAT 1SG friend ‘I sent the book to my friend.’ [Sranan]
The data presented above therefore suggests that a local relexification of the Gbe verby allative preposition did not take place in Sranan. What we do find in both (groups of) languages however, is the use of verb(id)s rather than dedicated prepositions for marking GOALs. In both Gbe and Sranan a verby rather than a prepositional strategy is therefore marshalled for the expression of GOAL-oriented motion. In some Gbe varieties the one-time V2 has progressed far enough along the grammaticalization chain to warrant being called a preposition while in others, the V2 retains verby characteristics. But in none of the Gbe varieties have the verbal origins of the GOAL-marking element been wholly obscured. This suggests that Sranan speakers could have modelled the realization of GOAL-oriented motion events on an originally verbal Gbe pattern. This tendency would have been reinforced by the existence of numerous lexicalized caused-motion and locomotion SVCs in Gbe that involve the use of full verbs in the pre-GROUND position, e.g. Ewe: tsɔ́ –vá [take–come] ‘bring’, kplɔ́ –yì [lead–go] ‘accompany’). The realization of ablative, i.e. SOURCE-oriented motion events is also characterized by minor differences between Sranan and Gbe. Consider the following example from Fon:
(27) ǹ sɔ́ àkwɛ́ ɖò gbàví ɔ̀ mɛ̀. 1SG take money LOC box DEF inside. ‘I took money out of the box.’ [Fon; Höftmann 1993: 140]
In the most common type of SOURCE-oriented motion description in Gbe, the general locative preposition (ɖò in the example above) marks the SOURCE and specifies the RELATION between FIGURE and GROUND. At the 148 Kofi Yakpo and Adrienne Bruyn same time, a postpositional locative noun (mɛ̀ ‘inside’ in the example above) expresses the REGION. What both types of locative elements share in semantic terms is that they do not contribute any directional meanings to the construction. Instead, both merely express PLACE notions. Since sɔ́ is a manner-of-motion verb rather than a directional verb, the PATH component of the motion event described in (27) above arises solely by implicature. Contrast this with example (28) from Ewe, which features the directional verb dò ‘exit’ and which contributes a PATH reference to the SOURCE-oriented motion event:
(28) lè ɣèmáɣì-á, nyè há̃ mè-dò lè sùkû xóxó. LOC that.time-DEF 1SG.EMP too 1SG-exit LOC school already ‘At that time, I too had already left school.’ [Ewe]
In the Gbe languages, the general locative preposition found in SOURCE-oriented constructions is formally identical with the locative-existential copula found as a predicator in PLACE-oriented constructions (cf. (10) above). Although derived from the copula, this from has been analyzed as a fully grammaticalized preposition in the Gbe languages when it occurs in locative constructions. Firstly, the form lè/ɖò is not normally marked for aspect or mood in structures like in (27) and (28). In addition, Aboh, Ameka, and Essegbey (2007) adduce evidence for the prepositional status of this form from the observation that a prepositional phrase introduced by lè can be fronted as a topic in a sentence like (29):
(29) lè aƒé-á mè lá, mè-kpɔ́ Kòfí LOC house-DEF inside TOP 1SG.SBJ-see NAME ‘IN THE HOUSE, I saw Kofi.’ [Ewe; Ameka et al. 2007: 9]
Fronting would not be possible if lè were a V2 in a serial verb construction as is the case with yì in (30) below (Ameka et al. 2007: 9):
(30) *yì àƒémè lá, mè-zɔ̀ . go home TOP 1SG.SBJ-walk ‘I WALKED home.’ [Ewe; Ameka et al 2007: 9]
The categorial status of the copula-derived locative preposition in Gbe is relevant with respect to the possibility of a local relexification of this form in Sranan. There is a substantial overlap in the functions of lè (Ewe, Gen, Aja) and ɖò (Fon) with Sranan na ‘LOC’. The fully prepositional status of lè The relexification of locative constructions in Sranan 149 and ɖò might therefore help to explain why the Gbe creators of Sranan did not select the corresponding Sranan locative copula de for prepositional functions – although they did select motion verbs whose meanings overlap with corresponding Gbe items in the expression of motion events (cf. (36)-(38) below): A SOURCE-oriented motion event involving the use of a preposition formally identical with the locative copula is complemented by another strategy in the Gbe languages. Alternatively, the directional ablative (i.e. SOURCE-oriented) verbs tsó (Ewe) and sín (Fon) ‘come from’ may be used to mark the SOURCE instead of the general locative preposition:
(31) e-tsɔ dziwui ɖeka tso e-ƒe mɔzɔɖaka me nɛ. 3SG-take shirt one (come)from 3SG-POSS suitcase inside DAT.3SG.OBJ ‘He took a shirt from his suitcase (and) gave it to him.’ [Ewe; Nyaku 1982: 47]
(32) ǹ sɔ́ àkwɛ́ sín gbàví ɔ̀ mɛ̀ . 1SG take money (come)from box DEF inside. ‘I took money out of the box.’ [Fon; Höftmann 1993: 140]
There appears to be a subtle difference between the alternatives in (27) and (32) above. When a SOURCE GROUND is introduced by the verbid tsó/sín rather than lè/ɖò the motion component of the event is emphasised and the event acquires a higher degree of dynamicity. The elements tsó (Ewe) and sín (Fon) are more fluid in their categorial status than the fully grammaticalized preposition lè/ɖò ‘LOC’. Firstly, the two forms may also be employed as common verbs with the meaning ‘come from’, as shown for Ewe in (33):
(33) Èʋè-dùkɔ́ ƒé àkpá ɖé, yé-wó ké tsó kéké Sudan. Ewe-nation POSS part INDF LOG-3PL EMP come.from EMP PLACE ‘A part of the Ewe nation, they even originate in far-away Sudan.’ [Ewe]
Secondly, these elements are characterized by a morphosyntactic behaviour suggestive of reduced verbiness (cf. Ansre 1966, Aboh, Ameka, and Essegbey 2007). Aboh, Ameka, and Essegbey (2007) show that 150 Kofi Yakpo and Adrienne Bruyn when Ewe tsó occurs in a locative construction, it may be optionally marked for the same aspect category as the preceding verb. When marked in this way, tsó is indistinguishable from the V2 of a common serial verb construction. When left unmarked, its distribution is similar to that of the fully grammaticalized locative prepositions lè/ɖò covered above:
(34) Kofi zɔ-na tsó (-ná) aƒéme ŋdí sía ŋdí NAME walk-HAB come.from(-HAB) home morning every morning ‘Kofi walks from home every morning’ [Ewe; Aboh, Ameka, and Essegbey 2007: 10]
Ablative motion events in Sranan are also instantiated in constructions bearing a strong resemblance to their Gbe counterparts (cf. Essegbey and Bruyn 2002). The SOURCE in Sranan SOURCE-oriented constructions is marked by means of a general locative preposition, just like in Gbe, namely the omnipresent (n)a ‘LOC’:
(35) mi teki a moni na (ini) a dosu (ini). 1SG take DEF.SG money LOC inside DEF.SG box inside ‘I took the money from the box.’ [Sranan]
The only notable difference between the Sranan and Gbe constructions is that the locative noun expressing the REGION (ini ‘inside’ in the example above) may once more be found either in a pre- GROUND or a post-GROUND position in Sranan – the alternatives are in parentheses. In addition, Sranan has the additional option of expressing SOURCE-oriented motion events through employing the directional verbs puru ‘remove’ or komoto/komopo ‘take out’ as V2s in argument-introducing SVCs. Sentences like (36)-(38) below are close Sranan equivalents to the Gbe constructions in (31) and (32) above. In the following three Sranan sentences, the locative noun ini ‘inside’ is again optional and may appear either in a pre- or a post-GROUND position. The possibility of a lexical choice between the near- synonyms puru, komoto and komopo shows that the structures below do not involve grammaticalized preposition(-like element)s in the V2 position, and that we are dealing with genuine serial verb constructions. Further, the SOURCE in the Sranan constructions is once more obligatorily marked with the general locative preposition na ‘LOC’: The relexification of locative constructions in Sranan 151
(36) mi teki a moni puru na a dosu (ini) 1SG take DEF.SG money remove LOC DEF.SG box inside ‘I took the money out of the box.’ [Sranan]
(37) mi teki a moni komoto na (ini) a dosu. 1SG take DEF money take.out LOC inside DEF.SG box ‘I took the money out of the box.’ [Sranan]
(38) mi teki a moni komopo na (ini) a dosu. 1SG take DEF money take.out LOC inside DEF.SG box ‘I took the money out of the box.’ [Sranan]
We have established that both Gbe and Sranan make use of complex locative constructions in which motion descriptions are jointly realized by verbs, prepositions and locative nouns. One reason for the participation of these different word classes in locative constructions lies in the scarcity of dedicated prepositions in Gbe, a typological feature that in fact characterizes the entire Niger-Congo phylum. This scarcity is made up for by the use of locative constructions ranging from more phrasal to more clausal structures. Indo-European, in turn, are typically phrasal. If we now direct attention towards GROUND-marking strategies in the three spatial relations of PLACE (essive), GOAL (allative) and SOURCE (ablative) we however see a significant difference between Sranan and Gbe. Sranan employs a unitary system, in which GROUNDS in the three relations are marked in the same way via na ‘LOC’, hence characterized by the pattern (PLACE/GOAL/SOURCE). Gbe, in contrast, features two alternatives: One is a bipartite system (PLACE/GOAL, SOURCE) in which PLACE and GOAL GROUNDS are marked in the same way (i.e. no pre-GROUND locative element). The other is a tripartite system (PLACE, GOAL, SOURCE) in which the GROUND in all three relations is marked by separate pre- GROUND elements (i.e. no locative element, ɖé/dò ‘ALL’ or lè/ɖò ‘LOC’/(t)só ‘ABL’). We can also establish that in typological terms, and disregarding the more recent contact-induced changes in Sranan, motion descriptions in Gbe and Sranan represent the serializing type: PATH (and RELATION) components of the motion event are exclusively expressed by verb(-string)s. In this, Sranan and Gbe differ from English and Dutch where PATH and Relation components are exclusively lexicalized in a preposition if a directional verb is absent. At the same time, it has also been shown that Gbe motion descriptions may be situated along a continuum with respect to the categorial status of the pre-GROUND element. While the elements ɖé/dò ‘reach; ALL’ (t)só ‘come from; ABL’ retain distributional characteristics peculiar to verbs, the element lè/ɖò behaves like a proper preposition when 152 Kofi Yakpo and Adrienne Bruyn it appears in a pre-GROUND position. Disregarding the obligatory use of a REGION element for the moment, the use of the PATH-denoting verbids ɖé/dò and (t)só as prepositions therefore represents a partial shift from the serializing type of locative construction towards the ‘prepositional’ pole of the continuum, in which a preposition rather than a verb expresses PATH. In comparison to Gbe, parts of the Sranan system represent a tidier form of the serializing type. For in Sranan, there are no half-way or fully grammaticalized locative prepositions with verbal origins. Instead, the V2s of a variety of conventionalized, “asymmetrical” SVCs (Aikhenvald 2006: 21) are recruited to express the locative meanings encoded by verby prepositions in Gbe (cf. ex. (25), (36)-(38). At the same time, we have seen that contact with Dutch is a pull factor in a similar movement towards the prepositional pole of the continuum. However, in Sranan the grammaticalization process is leading to the use of locative nouns in a pre-GROUND position rather than verbs, as in Gbe (cf. Sranan ex. (16)- (19). Despite these tendencies, Sranan and Gbe share a typological pattern in which SVC(-like) structures involving verbs or verbids may fulfil locative functions, where analogous structures in English and Dutch make exclusive use of prepositions.
3.3 From preposition in the lexifier to verb in Sranan
We now turn to a phenomenon that further corroborates the view that Sranan locative expressions are largely the outcome of pattern relexification. In the following, we will look at a set of five elements in Sranan whose etyma function as prepositions and locative particles in Dutch and English respectively. In Sranan, these elements are, however, multicategorial. On the one hand, they occur as prepositions or particles, as in their lexifiers. On the other hand, they are used as full verbs, and therefore appear in functions alien to those of the corresponding items in their lexifiers. We arrive at the conclusion that the presence of multicategorial verby prepositions in Gbe in the same syntactic position as these prepositions/particles in their lexifiers must have been the door-opener for the reanalysis of these elements into verbs in Early Sranan (cf. Bruyn 2008, 2009). As in the other cases treated so far, there is however no exact correspondence between Gbe and Sranan. The influences from Dutch are non- negligible and at the same time the independent development of some of these forms in Sranan must also be factored in. The relexification of locative constructions in Sranan 153
The five Sranan items contained in Table 2 are derived from English and Dutch prepositions, verbal particles and adverbs, hence non-verbal forms. In Sranan, these items however occur with verbal functions, while prepositional and particle uses are also attested. In what follows, we attempt to provide explanations for their behaviour:
Table 2. Multicategorial locative elements in Sranan Sranan item Verbal Non-verbal Lexifier etymon Lexifier word class meaning meaning doro ‘pass, arrive’ ‘through’ door ‘through’ (Du.) Preposition/particle romboto/ ‘surround’ --- roundabout (Eng.), Adverb/particle lomboto but also Gungbe lòbòtò ‘round’ lontu ‘surround’ ‘around’ rond ‘(a)round’ (Du.) Preposition/adverb abra ‘cross’ ‘over, across’ over (Eng.) Preposition/adverb opo ‘rise, raise’ ‘up, above’ up/op (Eng./Du.) Preposition/adverb
The individual Sranan forms in can be situated on a cline from top to bottom with respect to the degree of local relexification of Gbe forms. We will see that the two forms at the lower end (opo and abra) do not correspond to specific substrate forms. The uses of these forms nonetheless show the kind of incorporation into substrate derived structures that we have already observed with some of the locative nouns covered in the preceding sections. The first three forms in Table 2 manifest a close correspondence in terms of their semantic organization and morphosyntactic behaviour with corresponding Gbe forms. A point-by-point comparison between the Sranan and Gbe forms follows in Table 3. We exemplify the parallels between Sranan and Gbe by using the corresponding Ewe forms.
Table 3. Multicategorial locative elements in Ewe/Fon Ewe element Fon element Verbal meaning Spatial meaning ɖé dò ‘reach, arrive at’ ‘towards, to’ ƒò xlã lɛ́ lɛ́ dó ‘surround’ ‘round about’ tsò gbò ‘sever, separate, cut’ ‘across’
154 Kofi Yakpo and Adrienne Bruyn
The Sranan verb doro ‘pass (through), arrive’ in Table 6.3 is (phonologically) derived from the Dutch form door ‘through’. Dutch door is not used as a verb. It can be employed as a MEDIUM-denoting preposition as in door het boos lopen [through the forest walk] ‘walk through the forest’. It is also used as verb particle in more or less lexicalized collocations with more or less spatial meanings, i.e. door- kruisen [through-cross] ‘traverse’, door-leven [through-live] ‘live through (an experience)’. The following sentence shows the focal uses of Sranan doro as a telic GOAL-oriented motion verb with the meaning ‘arrive’:
(39) fa mi doro na oso, mi sisa lusu kaba. when 1SG arrive LOC house 1SG sister leave PRF ‘When I arrived at home, my sister had already left.’ [Sranan]
The case of doro in (39) above is a fine example for the reanalysis of an originally non-verbal form in Dutch into a verb in Sranan. We assume that this peculiar process of reanalysis was possible in this case and in the ones that follow below due to the multicategoriality of corresponding substrate items. The preceding section showed that the Gbe languages feature a grammaticalized allative preposition derived from a verb meaning ‘reach’ that can be employed as a lexical verb in some varieties. At the same time, Sranan was shown to employ lexicalized SVCs instead of the Sranan equivalent doro, albeit along a Gbe-type syntactic pattern (cf. §0). Beyond that doro has retained (or developed) semantic and syntactic characteristics of its Dutch prepositional source form. It is also used as a preposition and adverb/particle- like element with vague and metaphoric MEDIUM semantics in Dutch-influenced idioms like go doro [go through] ‘continue’ (< Du. doorgaan). A comparable situation holds with the Sranan forms lomboto/romboto ‘round about, around; surround’ and the near synonym lontu (< Du. ‘rond’/Eng. ‘(a)round’). These forms are presumably derived from English and Dutch etyma respectively and but the possibility of convergence with Gbe forms like lóbó(e) ‘round(ish)’ (Ewe) and lòbòtò/ròbòtò ‘round’ (Gun; Aboh, p.c.) should not be discarded. For one part, both forms are used as predicative nuclei in sentences like the following ones:
(40) a liba lomboto den oso. DEF.SG river surround DEF.PL house ‘The river flows around the houses.’ [Sranan] The relexification of locative constructions in Sranan 155
(41) den skowtu lontu a oso (…) DEF.PL police surround DEF.SG house ‘The police surrounded the house (…)’ [Sranan; Wilner 2007: 94]
Secondly, lomboto/romboto, just like doro above, is not normally employed as a PATH-denoting verby preposition/V2 in locative constructions like (42). Only lontu is accepted by our informants in a pre- GROUND position in a sentence like (43):
(42) *a liba e lon lomboto den oso. DEF.SG river IPFV run surround DEF.PL house ‘The river flows around the houses.’ [Sranan; field notes]
(43) a liba e lon lontu den oso DEF.SG river IPFV run surround/around DEF.PL house. ‘The river flows around the houses.’ [Sranan]
One possibility why the use of lomboto is rejected by our informants as a V2 in a structure like (42) above may be the contraction of the distributional potential of this item due to obsolescence – the form is classified as archaic by Wilner (2007). The form lomboto is already present in Schuhmann (1783) in examples such as the following - one of several, in which the form functions as a verb:
(44) meki wi rombotto hem SBJV 1PL surround 3SG.OBJ ‘Let’s surround him.’ [Sranan; Schuhmann 1783]
Contrary to lomboto, for which therefore have historical evidence for a verbal use, the categorial status of lontu in an example like (41) above is unclear. The form may simply be used as a preposition rather than a V2 in very much the same way as its English and Dutch cognate forms in clauses like the river flows around the houses/ de rivier stroomt rondom de huizen. Such multi-categoriality is also attested with doro, as well as with abra further below. An unequivocal example of a non-verbal, in this case adverbial use of lontu is given in (45):
(45) a luku lontu. 3SG IPFV around ‘He looked around.’ [Sranan; Blanker and Dubbeldam 2005: 127] 156 Kofi Yakpo and Adrienne Bruyn
Prepositional/adverbial uses of lomboto, as in (43) and (45) respectively, are not attested in our data. This is presumably so because the archaic form lomboto has retained its earlier uniquely verbal uses while lontu has acquired new non-verbal functions through contact with Dutch, probably reinforced by the phonological proximity of lontu and Dutch rond. A look at the Gbe substrate once more reveals close parallels with Sranan in the way functionally corresponding forms are used. The Fon form lɛ́ lɛ́ dó ‘surround’ appears as a finite verb preceded by a personal pronoun in (46). In (47), lɛ́ lɛ́ dó is found in a V2 slot in a structure that looks like an SVC, just like the corresponding Sranan form in (43):
(46) ye lɛ́ lɛ́ dó mì. 3PL surround 1SG.OBJ ‘They have surrounded me.’ [Fon; Höftmann 2003: 285]
(47) é dó kpá lɛ́ lɛ́ dó glè tɔ̀ n. 3SG put fence surround field 3SG.POSS ‘He put up a fence around his field.’ [Fon; Höftmann 2003: 285]
In the same vein, the Ewe expression ƒò xlã́ ‘surround’ (composed of the verb ƒò ‘beat’ and the inherent complement xlã́ ‘crookedness’) appears as a finite verb marked for habitual aspect in (48).
(48) wo-ƒo-a xla du-a kple fia la zi ɖeka. 3PL-beat-HAB crookedness town-DEF and chief DEF time one ‘They at once surround the town and the chief.’ [Ewe; Obianim 1990: 21]
In (49), appears in a pre-GROUND position similar to lontu in (43) above (disregarding the composite nature of the Ewe expression for ‘surround’). Hence irrespective of the categorial status that we may assign to lontu in (43) above, the surface structure of these constructions is similar in both languages.
(49) agbledela lá ɖe-a mɔ ƒo-a xla e-ƒe agble yeye. farmer DEF remove-HAB path beat-HAB crookedness 3SG-POSS farm new ‘The farmer clears a path around his new farm.’ [Ewe; Obianim 1990: 147] The relexification of locative constructions in Sranan 157
Further, in Ewe, just like in Sranan, the element denoting ‘surround’ is categorially ambivalent between verb and preposition. It is far less grammaticalized to prepositional status in Ewe than the verb-derived prepositions ɖé ‘ALL’, tsó ‘ABL’ and lè ‘LOC’. Evidence for this is provided by the optional use of TMA marking with ƒò xlã́ when it is found in the V2 position of an SVC, as evidenced by habitual aspect marking on both verbs present in the example above. So where the Sranan expression for ‘surround’ tends towards a prepositional status, and we suggest that this is through Dutch influence, the equivalent Ewe expression retains its verbal characteristics. A similarly complex case that once more shows the competition in Sranan between semantic and syntactic specifications likely to have been inherited from the substrates, internal development and Dutch influence is the case of abra, derived from English ‘over’ The item abra is employed as a main verb with the meaning ‘(to) cross’ in (50).
(50) wi o abra wan liba dyonsro. 1PL FUT cross one river soon ‘We are soon going to cross a river.’ [Sranan]
The item abra is also attested in Sranan in the collocation koti abra ‘(cut a)cross’, cf. (51). In the absence of further evidence, the could at first glance be analyzed as a lexicalized SVC in line with the analysis proposed for analogous structures involving komoto ‘go out, motion outward’ and go ‘go, motion toward’ (cf. (36) ff.). Just like example (43) involving lontu ‘around; sur(round)’ above, the structure may alternatively also be seen to involve a particle/adverbial use of abra:
(51) wi o koti wan liba abra dyonsro. 1PL FUT cut one river (a)cross soon ‘We are soon going to (cut a)cross a river.’ [Sranan]
However, it seems that an adverbial interpretation of abra is more convincing because the adjacency of koti and abra is not accepted by our informants when the GROUND is explicitly mentioned, as shown in (52):
(52) *wi o koti abra wan liba dyonsro. 1PL FUT cut (a)cross one river soon ‘We are soon going to (cut a)cross a river.’ [Sranan] 158 Kofi Yakpo and Adrienne Bruyn
Example (51) above shows a linear equivalence of constituents with the corresponding Dutch structure in (53) below, which involves the complex verb over-steken, composed of the particle over ‘over, across’ and the verb steken ‘jab’. In Dutch too, the adjacency of steken and over and hence a pre-GROUND position of over in these constructions is ungrammatical. Dutch influence on the semantics and the syntax of koti abra appears quite straightforward. We see this as supporting evidence for an adverbial interpretation of abra in these sentences.
(53) we steken de rivier over. 1PL jab:PRS:PL DEF river across ‘We are going to cross the river.’ [Dutch]
(54) *we steken over de rivier. 1PL jab:PRS:PL across DEF river ‘We are going to cross the river.’ [Dutch]
We also have adpositional uses of abra. We have evidence for historical uses of abra in a post-GROUND position in structures no different from the post-GROUND uses of locative nouns like tapu and ini, cf. (55):
(55) na mi hosso abra. LOC 1SG house across ‘across from my house’ [Sranan; Schuhmann 1783]
In modern Sranan, abra however seems to appear exclusively in a pre-GROUND position with a prepositional function:
(56) a opolangi frei abra a foto. DEF.SG plane fly over DEF.SG town ‘The plane flew over the town.’ [Sranan]
The relexification and contact scenario becomes even more intricate when we bring the corresponding English and Gbe structures into the picture. The equivalent English expression cut across is not only replicated structurally and semantically by the bi-composite structure of koti abra. However, a grammatical Sranan sentence can also be constructed without the adverb and with the “verb of crossing” alone. However, in English, the verb cut alone cannot be used for the act of crossing by itself. Compare the Sranan example in (57) and its English translation: The relexification of locative constructions in Sranan 159
(57) wi o koti wan liba dyonsro. 1PL FUT cut/cross one river soon ‘We are soon going to cross/cut across a river.’ [Sranan]
This peculiarity of Sranan can be explained by turning to the Gbe languages. In Gbe, the verb of crossing is equivalent to the verb ‘separate, sever, cut’. Hence we find the same Ewe verb tsò in (58) with a Patient and (59) with a GROUND object. In Fon the verb gbò may be used in the same two contexts (cf. Segurola and Rassinoux 2000: 220-21).
(58) amesi tso lã la ƒe ve la xɔ-a lã la ƒe kɔ. whoever cut animal DEF POSS throat TOP get-HAB animal DEF POSS neck ‘He who cuts the animal’s throat gets the animal’s neck.’ [Ewe; Obianim 1990: 166]
(59) esi wo-tso tɔsisi-a vɔ la, dzidzi ƒo Yakobo. when 3PL-cut river-DEF COMPL TOP pleasure hit NAME ‘When they had crossed the river, pleasure struck Yakobo.’ [Ewe; Nyaku 1982: 47]
In contrast to English and Dutch, however, Gbe features no lexicalized bi-composite structure equivalent to Sranan koti abra ‘cut across’ in the description of the crossing event. Distributional evidence suggests that tsò, like ƒò xlã́ ‘surround’ above, is not a grammaticalized preposition with the meaning ‘across’ either. The form retains verbal properties regardless of its syntactic position. Hence a scenario is plausible in which a Gbe verb for ‘cut’ was relexified in Sranan to encompass the meaning of ‘cross’. At the same time, a carry-over from English and contact with Dutch and would have encouraged the retention of abra with an adverbial function. This might have initially been limited to the act of ‘crossing’ in the collocation koti abra. However, the existence of other multicategorial elements with adpositional and verbal characteristics in Sranan and Gbe would have facilitated the extension of abra to verbal functions not found in English and Dutch. This once more shows how competing substrate, lexifier and superstrate forces have produced a versatile, multicategorial item like abra with its wide range of verbal, adverbial and adpositional uses. The final element in Table 6.2 above is opo, a form that is already attested in earlier stages of Sranan with the verbal meanings of ‘rise’ and ‘raise’. 160 Kofi Yakpo and Adrienne Bruyn
(60) effi ju srefi no kann hoppo, mi sa hoppo ju. if 2SG self NEG can rise 1SG FUT raise 2SG ‘If you yourself cannot get up, I’ll get you up’ [Schumann 1783]
Such intransitive and transitive uses of opo are still commonplace in contemporary Sranan, an example for the latter use is given in (61):
(61) op gegeven moment wan hei opo en ede. on given moment one agouti raise 3SG head ‘Suddenly an agouti raised its head.’ [Sranan]
The form opo has no exact equivalent in Gbe, where separate lexemes and expressions cover the intransitive and transitive notions respectively. In Ewe, for example, we find fɔ́̃ ‘get up’ and the SVC kɔ́ – yì dzĭ [take – go – upper surface] ‘raise’. In this respect, these uses of opo in Sranan are innovative vis-à-vis both English and Gbe. Still, the preceding discussion has shown that the reanalysis of the source language preposition up/op to a verb in Sranan, proceeded along the same path as that of the other four forms. The remarkable aspect of this trajectory is that the forms have retained (or (re-)acquired through contact with Dutch) at least some of the prepositional/particle uses of their source language etymons. Within the broader scenario of relexification, here too, the Gbe languages seem to have provided the blueprint for the reanalysis of individual forms and of morphosyntactic patterns. In this vein, a local relexification may be seen to have led to the overlap in semantics and morphosyntactic behaviour between a Sranan form like lontu ‘surround’ and Ewe fò xlã́ in examples like (41) and (49) above. At the same time, the development of the verbal uses of a form like abra ‘(to) cross’ can only be seen within a more generous relexification perspective. This view includes the possibility of a carry-over of morphosyntactic specifications and functions, hence patterns, without necessarily involving a one-to- one mapping of an individual phonological shape in the lexifier language with a specific lexical item in the substrate (cf. Bruyn 2008, 2009). The following section explores this possibility further and attempts to find explanations for the large degree of variation encountered in Sranan. The relexification of locative constructions in Sranan 161
4. Towards an explanation of variation in Sranan
Previous sections have shown that Sranan locative constructions are characterized by quite a high degree of morphosyntactic and functional variation of the participating elements. In this section, we will suggest that besides influence from Dutch, and transfer from the lexifier English in earlier stages of Sranan, the cause of variation can also be sought in the equally broad variety of constructions found across the substrates of Sranan. In this respect, the situation in the other (group of) substrate(s) is relevant, namely the Kikongo cluster and closely related languages like Kimbundu, which have been shown to constitute the second most important group of substrates next to Gbe (cf. e.g. Arends, Kouwenberg, and Smith 1995; Huttar 1986). The first characteristic in need of explanation concerns the variation encountered in the use of pre- GROUND and post-GROUND locative nouns in Sranan. Sranan structures in which the locative noun is found in a pre-GROUND position may have been influenced by substrate structures just as much as by the lexifier English and the superstrate Dutch. In Ewe, some locative nouns may also appear in a pre-GROUND position. When used in this way, the locative noun is linked to the GROUND via the dative marker ná, derived from a verb meaning ‘give’. In such instances, the dative marker may in fact be likened to a possessive linker. The following examples show both alternatives in Ewe:
(62) Àfí lè ŋgɔ́ ná Kòfí. NAME be.at:PRS front DAT NAME ‘Afi is in front of Kofi.’ [Ewe]
(63) Àfí lè Kòfí ŋgɔ́ . NAME be.at:PRS NAME front ‘Afi is in front of Kofi.’ [Ewe]
Kikongo locative nouns canonically appear in a pre-GROUND position, a feature common to the Narrow Bantu branch of the Niger-Congo phylum. Beyond that, the language features the typical bipartite structure of locative expressions encountered throughout Niger-Congo, and as an areal feature, far beyond (e.g. in the Chadic language Zina Kotoko, c.f. Aboh 2010). Hence in (64), there is a general locative element ku ‘LOC’ (generally a noun class prefix in the Bantu languages, but written separately from the noun in some of the sources consulted). There is also a locative 162 Kofi Yakpo and Adrienne Bruyn noun indicating the region, namely ntundu ‘top’, as well as a possessive element a ‘POSS’ that links these locative elements to the GROUND:
(64) e mpu ame iina ku ntundu a meza. DEF hat 1SG.POSS BE.AT LOC top POSS table ‘My hat is (lying) on the table.’ [Kikongo; Tavares 1915: 80]
Kimbundu, an immediate relative of Kikongo spoken in Northern Angola, features analogous locative constructions. The example in (65) involves the use of a general locative element (the noun class prefix bu ‘LOC’) and a locative noun (kanga ‘outside’), which invariably appears in a pre-GROUND position. Just like in the Kikongo and Gbe examples above the locative noun is linked to the GROUND noun via an intervening element, in this case the possessive linker ria ‘POSS’ (which concords with the noun class of the preceding head noun):
(65) o sanzala ietu a-i-tung-u bu kanga ria muxitu. DEF village our 3PL.SBJ-3SG.OBJ-build-PFV LOC outside POSS forest ‘Our village is built outside of the forest.’ Lit. ‘Our village, they built it outside of the forest.’ [Kimbundu; Chatelain 1888: 116]
In Kikongo and Kimbundu, we therefore find constructions with a surface structure similar to Sranan locative constructions involving pre-GROUND locative nouns, except that Sranan has no prefixes. The only element “missing” to make Sranan structures like (37) above virtually isomorphic with the Bantu and the Gbe structures covered above is the possessive linker. We will return to this aspect in due course. A characteristic that sets Sranan apart from Gbe is the categorical use of the general locative preposition na ‘LOC’ before all types of GROUNDs. We saw in (2)-(5) above that the preposition appears before GROUNDs with a PLACE, a GOAL and next to the preposition fu, a SOURCE role. We have shown that the first and second of these three participant roles are not marked by the corresponding general locative preposition in Gbe (cf. (21) above). How can the more extensive participant marking functions of the Sranan general locative preposition na be explained? We do not reject the explanation by Essegbey (2005: 256) that leveling (“generalization” in the author’s terms) may have at least contributed to the crystallization of Sranan na into an obligatory locative marker. But we will also go on to show that the origins The relexification of locative constructions in Sranan 163 of the obligatory presence of na in Sranan locative adjuncts may as well lie in corresponding ones in the substrate. Bantu locative constructions are characterized by the use of general locative elements in a pre- GROUND position. These elements have a similarly broad range of functions as Sranan na ‘LOC’. Also relevant in this context is that the corresponding Bantu locative prepositions are, just like na in Sranan, normally not derived from verbs, at least not in a synchronically transparent way. Three sentences follow that exemplify the use of the general locative element ku with a PLACE (66), a GOAL (67) and a SOURCE (68) in Kikongo:
(66) ku Matadi tuamonana (...) LOC PLACE see:RECP:PST:1PL ‘In Matadi we saw each other (…)’ [Kikongo; Söderberg and Widman 1966: 57]
(67) ku Kisantu kayele. LOC PLACE go:PST.HST:3SG ‘He went to Kisantu.’ [Kikongo; Anonymous 1964: 37]
(68) ntama yâkatuka ku bwâla dyâme. since.long leave:PST:1SG LOC village 1SG.POSS ‘It’s a long time since I left my village.’ [Kikongo; Dereau 1955: 138]
The Bantu languages in general do not employ the kind of prototypical SVCs that we have seen in Gbe. Nevertheless it has been observed that structures reminiscent of SVCs are specifically employed to express PATH throughout Niger-Congo, whether a language is serializing or not (e.g. Creissels et al. 2008: 146). This allows the conclusion that the relexification of PATH patterns in Sranan involving motion-verbs like komoto ‘come out’, puru ‘remove’ and go ‘go’ may have been modelled not only on Gbe, but also on Bantu. And indeed, we also find a verb-derived element expressing (SOURCE-oriented) PATH in Kikongo, namely the verb –tuka ‘come from’. The use of katuka ‘leave’ as a finite lexical verb can be seen in (68) above. The following example shows the use of –tuka as a verbid preposition with the meaning ‘from’, once more in combination with the general locative preposition:
(69) tuka ku Matadi nate ye Leopoldville. (come)from LOC PLACE until with PLACE ‘From Matadi to Leopoldville.’ [Kikongo; Söderberg and Widman 1966: 61] 164 Kofi Yakpo and Adrienne Bruyn
What remains to be explained at this point is the absence in Sranan of a possessive linker in the locative constructions involving pre-GROUND locative nouns that we have seen so far (cf. e.g. (16)-(19) above). An explanation is required because Sranan possessive/modification structures involving full nouns either have the constituent order [Possessor–Possessed] as in Pieter oso ‘Pieter’s house’ or they feature the inverse order with an intervening possessive linker, namely the associative preposition fu, hence [Possessed–fu–Possessor] as in a oso fu Pieter. The order [Possessed–Possessor] encountered in locative constructions with a pre-GROUND locative noun like na ini a dosu [LOC inside DEF box] is therefore not encountered elsewhere in the language. The answer may be found in Early Sranan as well as in Modern Sranan. In Modern Sranan, we sometimes find locative constructions featuring the possessive linker fu between the locative noun and the GROUND, as in the following example:
(70) na fesi/tapu fu a skowtu-oso LOC front/top POSS DEF.SG police-house ‘to an area in front/above the police-station’ [Sranan; Norval Smith, p.c., data provided by Lilian Adamson]
The Sranan speakers consulted see a slight semantic difference between structures involving the possessive linker fu ‘of’ as in (70), and those without them (cf. examples (16)-(19) above). Structures with the possessive linker are seen as more “literal” and “emphatic” in their spatial meaning (Hein Eersel, p.c.), and the translation of (70) provided by Lilian Adamson suggests a more specific meaning of these structures as well. It seems, however, that structures involving the possessive linker were less specialized in their meaning in Early Sranan, and could have constituted a regular means of forming locative constructions, cf. the following example:
(71) na inni va wi hatti. LOC inside POSS 1PL heart ‘in our hearts.’ [Sranan; Schuhmann 1783]
It is from these kinds of overt possessive structures in Sranan, that the pre-GROUND use of locative nouns as in (16)-(19) above without a linking element could have developed and been conventionalized, presumably already reinforced at an early stage by English and Dutch prepositional structures. The optionality and marginal use of the linker in Modern Sranan will The relexification of locative constructions in Sranan 165 have also facilitated the development of purely prepositional functions of locative nouns in recent times. One conclusion that can be drawn from the discussion so far is that the input into Sranan could have been highly varied from the very beginning. In fact there is quite a degree of morphosyntactic diversity already present within and between the Gbe languages themselves. What Sranan, Gbe and Bantu share is the scarcity of Indo-European style prepositions and the corresponding use of bipartite locative structures involving a general locative preposition, locative nouns, and to some extent, verby PATH- denoting locative elements. We therefore concur with other accounts claiming that the underlying typological unity of the African input into Sranan and the Afro-Caribbean Creoles in general facilitated the transfer of substrate features (cf. e.g. Alleyne 1980; Faraclas 1987; Singler 1988). In addition, while Gbe has no exact equivalent of the Sranan general locative preposition, we do find a functionally identical form in Kikongo and a closely related language like Kimbundu. It is therefore not necessary to look at English and Dutch influence as the primary sources of the variation in the pre- and post-GROUND position of locative nouns. Yet, these two languages have of course contributed to the structural and semantic diversification of Sranan locative expressions. The next section will show that Sranan in fact maximizes the possibilities inherited from various sources.
5. Summary of findings and discussion
We can now summarize the features of the entities that we have been referring to as “patterns”. A pattern has been shown to consist of a systematic functional-semantic and morphosyntactic relation of at least two forms with each other. A pattern therefore includes specifications of relational features. In the following, we attempt to classify these features according to their possible origins in the input or contact languages of Sranan. We arrive at the conclusion that Gbe and Kikongo provided the patterns for the majority of the semantic and morphosyntactic features of locative constructions in Sranan. That said, locative constructions in contemporary Sranan are nevertheless the outcome of a complex interaction of substrate and superstrate (i.e. Dutch) patterns. As a result, Sranan manifests an unusual richness in the expressive possibilities of locative relations. The characteristics of locative constructions in Sranan and their relation vis-à-vis constructions in the other relevant languages are summed up in the following ten points: 166 Kofi Yakpo and Adrienne Bruyn
1. Many Sranan locative constructions are complex syntactic structures that may be seen to involve two interlocked dependency relations: a general locative preposition introduces a prepositional phrase; within the PP in turn, a locative noun functions as the head and possessed noun in a possessive relation, with the
dependent, the GROUND, functioning as a possessor noun. 2. Sranan, Gbe and Kikongo all have a distinct locative copula and only few fully grammaticalized locative prepositions.
3. Sranan, Gbe and Kikongo have a general locative preposition which may introduce participants with PLACE
and SOURCE roles. In Sranan, the preposition also introduces participants with GOAL and PATH roles. In contrast to Gbe, the Sranan locative preposition is not transparently derived from a verb in the language.
4. Sranan, Gbe and Kikongo employ locative nouns denoting a REGION. In Gbe, these are mostly found in a post-
GROUND position but a pre-GROUND is also possible. In Kikongo, only a pre-GROUND position is attested. In
Sranan both a pre- and a post-GROUND position are possible. 5. The use of locative nouns is not obligatory in Sranan. While Gbe locative constructions only dispense with locative nouns in the context of referential specificity, corresponding Sranan and Kikongo constructions are grammatical and self-contained through the use of the general locative preposition alone. 6. While Gbe locative nouns differ quite a lot in their degree of nouniness, all Sranan locative nouns covered appear to occupy more or less the same intermediary position between noun and adposition.
7. In Gbe, the GOAL can be realized as a transitive object or be introduced by a verby allative preposition. In
Sranan and Kikongo, GOALs in both types of motion events are introduced by the locative preposition. In
Sranan, Kikongo and Gbe, SOURCE is marked overtly and neither the general locative preposition nor locative nouns contribute any directional meanings. 8. However, the picture becomes more complex in both Sranan and Gbe when ongoing grammaticalization is
taken into account. In some Gbe varieties the pre-GROUND locative elements tsó/sín and ɖé/dò appear to have completed the transition to full prepositions. In such varieties, these grammaticalized prepositions encode
PATH information in the same way as the English prepositions out of/from and to(ward). However, REGION is still expressed separately in a locative noun. 9. In Sranan, the grammaticalization path towards prepositional status has taken the opposite route. Contact with Dutch has led to the rise of European-style locative structures: locative nouns now overwhelmingly in the pre-
GROUND position while the general locative preposition na and a locative noun denoting REGION are omitted. This development has led to an isomorphism of Sranan and Dutch patterns, but also to the partial overlap of contemporary Sranan/English/Dutch and Gbe patterns. 10. Sranan also features a small set of items derived from lexifier prepositions/adverbials that have acquired verbal uses. These elements retain their prepositional functions in Sranan and may also be used adverbially, like in Dutch and English. The relexification of locative constructions in Sranan 167
Table 4 compares the features of locative constructions in Sranan and the African and European input languages. The table also implicitly provides hypotheses about the origins of each of these features in Sranan as a consequence of substrate transfer (also present in Gbe and Kikongo), lexifier transfer (also present in English), superstrate transfer (also present Dutch) and internal development (not attested in any of the four input languages, indicated by N.A. in the table header). Note that the term “preposition” refers to the kind of verby preposition that we have seen to be characteristic for Gbe, as well as the unicategorial type of preposition that we find in English and Dutch:
Table 4. Comparison of features of locative constructions Features No. Sranan Gbe Kikongo English/Dutch N.A.
LOC copula expresses SPATIAL 1 + + + RELATION General LOC preposition 2 + + + expresses SPATIAL RELATION Locative nouns denote REGION 3 + + + Prepositions express PATH & 4 + + + SPATIAL RELATION only Prepositions express PATH, 5 + + SPATIAL RELATION & REGION Some prepositions function as 6 + + verb particles General LOC preposition marks 7 + + PLACE & GOAL General LOC preposition marks 8 + + + SOURCE POSS marker may link locative 9 + + + + element & GROUND Relatively open class of V2s 10 + + may mark GOAL & PATH
On the basis of the characteristics enumerated in points 1-10 above, and the summary of functions in Table 4, we can establish the following: Seen from the perspective of Gbe, the principal source of relexified patterns in Sranan, we find features in Sranan that represent the workings of centripetal and centrifugal forces. Centripetal forces manifest themselves where the functions of Sranan elements are coterminous with those found in Gbe. 168 Kofi Yakpo and Adrienne Bruyn
Hence we would expect Gbe substrate patterns to have driven the emergence of 1–4 and 8. Centrifugal forces show themselves in features not found in Gbe, hence 5–7. Among these, features 5 and 6 reflect the dominance of transfer from the superstrate Dutch, and potentially also lexifier transfer from English. Feature 7 is particularly interesting because it represents the only instance where the two substrates diverge. Hence with respect to the functions of locative elements there is a near-complete overlap between Gbe and Kikongo. Yet, we also pointed out earlier that internal development may have also contributed to the existence of feature 7. In between these two poles, feature 10 seems to represent a case of internal development, albeit closely modelled along the Gbe pattern in which SVCs rather than “pure” prepositions express the PATH component in a caused-motion event. Finally, feature 9 may represent a case of substrate, lexifier and superstrate convergence, since all input languages may potentially employ structures involving a possessive linker. One conclusion to be drawn from Table 6.4 is that all types of locative constructions, save two (i.e. the Dutch-influenced SPATIAL RELATION/PATH/REGION conflated prepositional phrase and the use of verbal particles) encountered in Sranan can be accounted for by appealing to corresponding structures within the Gbe and Kikongo substrates alone, either fully or in part (i.e. feature 10). The constituent order (from left to right) and structure of the Sranan, Gbe and Germanic (English and Dutch) locative constructions covered in this chapter are represented schematically Table 5 below. The following abbreviations hold for the headers of the slots: No = construction number; PLOC = general locative or other preposition; NLOC1 = locative noun slot 1; Linker = (possessive or other) linker; GROUND = GROUND; NLOC2 = locative noun slot 2:
Table 5. Structure of locative constructions
Language(s) No PLoc NLoc1 Linker GROUND NLoc2 Sranan 1 + + (+) + 2 + + + 3 + + Gbe 4 + + + 5 + + + + Kikongo 6 + + + + English/Dutch 7 + (+) +
The relexification of locative constructions in Sranan 169
Table 5 shows an overlap between Sranan, Gbe and Bantu patterns (1), (5) and (6), bearing in mind that the use of the (possessive) linker in (1) represents a semantically more specialized type of locative construction, and is rather rare in Modern Sranan, hence (+). The same notation principle applies for (7), where the use of a linker in English structures like on top of the mountain are more specific than on the mountain as well, hence (+) in the linker column. An overlap between Sranan and Gbe alone exists with respect to patterns (2) and (4), and between Sranan and Germanic alone with respect to patterns (3) and (7). Modern Sranan is therefore characterized by a maximal number of options with respect to the number of available patterns. In fact, Sranan unites all the possibilities found in the substrates, the lexifier and the superstrate. Even if we know little about the relative frequency of each pattern in Sranan, the combined result of substrate retention from Gbe and Bantu, lexifier retention from English, superstratal contact with Dutch as well as internal development gives Sranan speakers an unusually large range of options in the expression of locative relations. A final issue to be addressed is the nature of relexification. In the preceding sections, we have argued that certain types of Sranan locative constructions represent instances of pattern relexification, hence of clusters of items, rather than of individual items alone. In our case, the cluster that constitutes a Sranan locative construction forms a syntactic category, composed of a string of morphemes, which in turn, enter into syntactic relations with each other. In the Gbe-like Sranan structures that we have seen so far, these syntactic combination rules involve nested dependency relations: The adjunct PP is headed by the general locative preposition na ‘LOC’, while the GROUND contained in the PP is a dependent of the locative noun. In accordance with its formal complexity, the entire structure may also be seen to have a more complex meaning than a corresponding single item: It contains information on the SPATIAL RELATION, the GROUND and the REGION. This contrasts with the meaning of a single relexified item like tapu, which, taken by itself, only conveys information on a particular SPATIAL RELATION. An individual item nevertheless has a special role to play in pattern relexification. In fact, a semantic matching, the creation of a “conceptual link” (cf. e.g. Heine and Kuteva 2010: 89) between an individual English(-derived) and a Gbe item must have constituted the basis for the development of semantically and formally more complex structures modelled on Gbe. The interlingual identification of the English item top with a Gbe (Fon) item like jí ‘top’ would have occurred on the basis of shared meaning. In this particular case the match is quite close. Both forms not only designate a superior 170 Kofi Yakpo and Adrienne Bruyn location, their meaning also includes contact with the GROUND. At the same time, the nouniness of top in English must have provided further matching opportunities for Gbe speakers. Therefore individual items must have functioned as “pivots” (Matras 2009: 240-42) during the calquing process of Gbe locative structures. Such pivots entail the occurrence of other forms plus their combinatoric possibilities, hence relational structures. This include morphosyntactic specifications when individual lexical items belong to a paradigmatic class. Thus a superior location designated by tapu ‘top’ requires an inferior location, expressed by ondro ‘bottom’ and so forth. Much of the available evidence suggests that speakers want to emulate underlying semantic relations in the recipient language and that this need drives the recruitment of the corresponding phonological material and morphosyntactic structures in the recipient language (cf. Heine and Kuteva 2010). In the case of tapu ‘top’, the semantic relation is a SUPERIOR PLACE RELATION INVOLVING CONTACT. Further, it is one that requires the presence of elements expressing a FIGURE, a SPATIAL RELATION, a GROUND and a REGION. This process of matching and extension can be represented schematically, as in Figure 1.
Figure 1. Pattern relexification
ondro ini baka
Lexis Paradigmatic jí relations
Identification tapu Dependency relations top Syntagmatic relations Phonology
Constituent Distributional order potential
The relexification of locative constructions in Sranan 171
In the scenario presented in Figure 6.1, the relexification of the individual item as a point of departure to pattern relexification is not only crucial because of interlingual identification. Local relexification may also transfer “embryonic” syntactic information along with lexical one. Hence the specification of the syntactic category of an item like tapu as (a type of) a noun could be seen to be part of the lexical entry. But this already provides some distributional information about tapu, for example its potential to cooccur with another noun within the same NP. Meanwhile, pattern relexification could be seen to carry over specifications that are more syntactic in nature along the lexicon-syntax cline: constituent structure and order, e.g. the pre- or post-GROUND position of tapu or its direct adjacency vs. the presence of an intervening linker or the dependency relation holding between the head noun tapu and the GROUND noun. Pattern relexification also allows for minor morphosyntactic differences between Sranan and its main substrate Gbe, e.g. the rather balanced use of pre- and post-GROUND structures, and the obligatory presence of the locative preposition. These changes in constituent order do not affect the nature of the dependency relation between the locative noun tapu and the GROUND, nor the REGION function of tapu. Likewise, the generalized use of the locative preposition in Sranan, in contexts where Gbe makes no use of it, is not as contradictory as it may seem. This is because an overarching feature of the general locative prepositions in Sranan, Kikongo and Gbe is their potential to mark GROUNDs in motion events, rather than in static events alone. The respective Sranan and Kikongo locative prepositions apply this function indiscriminately to SOURCE and GOAL GROUNDs. In contrast, Gbe limits the use of the preposition to SOURCE GROUNDs. In that, Gbe displays a typologically common pattern, in which PLACE and GOAL are marked in a unitary fashion, while SOURCE is marked in a different way (cf. Creissels 2006).
6. Concluding remarks
We have seen that spatial relations in contemporary Sranan are expressed through a broad range of constructions. Some of these quite clearly reflect the historically more recent influence of the Dutch superstrate. The original system however clearly reflects the influence of the substrate languages of Sranan. These “Niger-Congo” structures are markedly different from equivalent “Indo-European” ones. They are bipartite, hence feature two functionally distinct locative elements, namely a general locative preposition and a locative noun. In semantic terms, these structures overtly encode 172 Kofi Yakpo and Adrienne Bruyn two spatial notions separately: While the locative preposition places FIGURE and GROUND in a spatial relation to each other in a general manner, the REGION element provides specific information about the space attached to the GROUND. In the Indo-European type of locative construction, we may find conceptually similar spatial descriptions in locative constructions like at the top of the building and at the back of the car. These constructions also feature a fairly general preposition (at) and a REGION element (back). But we have noted a profound typological rift between Sranan, Gbe and Kikongo on the one hand, and English and Dutch on the other, in the way these bipartite structures are employed in the description of motion events. In English and Dutch, prepositions whose meanings include a PATH component are employed when motion events are described. Hence an English clause like he removed it at the back of the car cannot be interpreted to involve a SOURCE-oriented motion. In contrast, Sranan and its substrates have the option of describing certain types of motion events in exactly this way because these languages offer the possibility of expressing the motion component through verbs alone. At the same time, the bipartite structure featuring the general locative preposition with its static semantics is maintained. Although a deeper analysis of these structures in Gbe and Kikongo revealed subtle differences between the two, the general constellation was found to be characteristic of languages from other branches of Niger-Congo as well. The rather isomorphic nature of locative constructions in Niger- Congo contrasts with the portmanteau prepositions characteristic of Indo-European, which conflate Relation, PATH and REGION (cf. Talmy 1985, 2000). The large number and manifold origins of locative prepositions in Indo-European languages tallies with the heavy functional load of this word class. In English, for example, we find denominal forms like (in) front (of) and down, adverb-noun combinations like inside, outside and (a)midst, and deverbal elements like past. With respect to the expression of spatial relations, Sranan is typologically unusual. The language appears to allow the use of the entire range of locative structures encountered in the substrate languages, the lexifier and the superstrate. At the same time, the existence of locative constructions with “Niger- Congo” semantics and a Gbe constituent order in particular, quite clearly points to relexification. We identified pattern relexification as the cause of the wholesale carry-over of substrate semantics plus morphosyntactic specifications into Sranan. This approach also makes allowance for minor differences in the functions and behaviour of Gbe and Sranan items. The distinction between the transfer of lexical material per se on the one hand, and the transfer of lexical and morphosyntactic properties on the The relexification of locative constructions in Sranan 173 other, is of course not new in the study of language contact. The latter phenomenon has been referred to in the literature (with varying degrees of overlap in meaning) by terms like “calquing” (Haugen 1950), “metatypy” (Ross 1996, 2001), “pattern transfer” (Heath 1984), “grammatical replication” (Heine and Kuteva 2003, 2005, 2010), “pattern replication” (Matras 2009), “rule borrowing” (Boretzky 1993), “apparent grammaticalization” (Bruyn 1996) and last but not least “relexification (of patterns)” (e.g. Lefebvre 1993, 1998; Lumsden 1999; Migge 2003; Muysken 1981, 1997; Voorhoeve 1973). Beyond providing more evidence for the reality of the process of pattern relexification, we hope to have additionally shown that the concept of pattern transfer can be used as an analytical tool to describe contact effects that have involved language creation, rather than mere convergence between existing systems. Secondly, a careful areal-typological analysis of the corresponding substrate structures can strengthen the case for substrate transfer and relexification in a “new” language like Sranan. In the absence of such an analysis, certain features (e.g. the general GROUND-marking function of na ‘LOC’ in Sranan) may otherwise be prematurely attributed to internal development. Thirdly, we may well encounter a diversity in Sranan locative constructions that appears bewildering at first sight. But we have seen that much of this apparent diversity is superficial in nature, for it chiefly concerns constituent order. In contrast, morphosyntactic relations like the nature of dependency, as well as the semantic structure of spatial descriptions remain highly similar in Sranan and the substrates. We hope to have shown that these contact-induced similarities are so systematic and profound that they may be seen as yet another manifestation of the Transatlantic Sprachbund that unites the West African coastal belt with the Caribbean.
Abbreviations: ó = high tone; ò = low tone; ō = mid tone; ABL = ablative; ADV = adverbial; AFF = affix; ASS = general associative preposition; CL = noun class prefix; COM = comitative preposition; COMP = complementizer; COMPL = completive aspect marker; COP = locative-existential copula; DEF = definite article; DP = discourse particle; Du. = Dutch; Eng. = English; HAB = habitual aspect marker; HST = hesternal; INDF = indefinite article; INF = infinitive; INT = intensifier; IPFV = imperfective aspect; ITI = itive particle; LOC = general locative preposition; LOG = logophoric pronoun; NLOC = ‘nouny’ locative element; m = masculine; NEG = negator; NOM = nominalizer; P = postposition; PASS = passive; Ploc = locative preposition;
174 Kofi Yakpo and Adrienne Bruyn
POSS = possessive; POT = potential mood marker; PRF = perfect marker; PRS = present tense; PST = past tense marker; Q = question particle; SBJV = subjunctive complementizer; SD = sudden discovery and narrative tense;
SVC = serial verb construction; TOP = topic; 3SG = third person singular; VEN = ventive particle; Vloc = ‘verby’ locative element; V1 = initial verb in an SVC; V2 = second verb in an SVC
References
Aboh, Enoch A (2010) The P-route. In Cinque, Guglielmo and Luigi Rizzi (eds) Mapping Spatial PPs: The Cartography of Syntactic Structures Vol. 6. New York: Oxford University Press. 225-260. Aboh, Enoch A. (2005) The category P: The Kwa Paradox. Linguistic Analysis 32: 615-46. Aikhenvald, Alexandra Y (2006) Serial verb constructions in typological perspective. In Aikhenvald, Alexandra Y. (ed) Serial verb constructions. Oxford: Oxford University Press, 1-68. Alleyne, Mervyn C (1980) Comparative Afro-American: An historical-comparative study of English-based Afro- American dialects of the New World. Ann Arbor: MI. Ambouroue, Odette (2007) Eléments de description de l’orungu, langue bantu du Gabon (B11b). PhD dissertation, Université Libre de Bruxelles. Ameka, Felix K. (1995) The linguistic construction of space in Ewe. Cognitive Linguistics 6: 139-181. Ameka, Felix K. (2003) Prepositions and postpositions in Ewe (Gbe): empirical and theoretical considerations. In Zribi-Hertz, Anne and Patrick Sauzet (eds) Typologie des langues d’Afrique & universaux de la grammaire, vol. ii. Paris: Harmattan, 41-67. Ameka, Felix K. and James Essegbey (2001) The expression of complex translational motion events in three verb-serializing languages. In Kelly, Ann and Alissa Melinger (ed) Annual report 2001. Nijmegen: Max Planck Institute for Psycholinguistics, 94–97. Ameka, Felix K. and James Essegbey (2006) Elements of the Ewe grammar of space. In Levinson, Stephen C. and David P. Wilkins (eds) Grammars of space: cognitive explorations, Cambridge: Cambridge University Press, 359–399. Ameka, Felix K. and Stephen C. Levinson (2007) The typology and semantics of locative predicates: posturals, positionals, and other beasts. Linguistics 45–5/6, 847-871. Aboh, Enoch, Felix K. Ameka and James Essegbey (2007) Moving from verbs to prepositions in Gbe (West Africa). Unpublished draft paper. Anonymous (1964) Leçons de kikongo par des bakongo. Eegenhoven -- Louvain. Ansre, Gilbert (1966) The verbid – a caveat to “serial verbs”. Journal of West African Languages 3, 29-32. Arends, Jacques (1995) Demographic factors in the formation of Sranan. In Arends, Jacques (ed) The early stages of creolization. Amsterdam: John Benjamins Publishing, 233-285. Arends, Jacques, Silvia Kouwenberg and Norval Smith (1995) Theories focusing on the non-European Input. In Arends, Jacques, Pieter Muysken and Smith, Norval (eds) Pidgins and creoles: An introduction. Amsterdam: Benjamins. Blanker, J.C.M & J. Dubbeldam (2005) Woordenboek Sranantongo. Utrecht: Prisma. Boretzky, Norbert (1993) The concept of rule, rule borrowing, and substrate influence in creole languages. In Mufwene, Salikoko (ed) Africanisms in Afro-American language varieties. Athens: University of Georgia
Press. Bruyn, Adrienne (1995) Grammaticalization in creoles: the development of determiners and relative clauses in Sranan. Amsterdam: IFOTT. Bruyn, Adrienne (1996) On identifying instances of grammaticalization in creole languages. In Baker, Philip and Anand Syea (eds) Changing meanings, changing functions: papers related to grammaticalization in contact languages. London: University of Westminster Press, 29-46. Bruyn, Adrienne (2003) Grammaticalization, réanalyse et influence substratique: quelques cas du Sranan. In Kriegel, Sibylle (ed) Grammaticalization et réanalyse: approches de la variation créole et française. Paris: CNRS Éditions, 25-47. Bruyn, Adrienne (2008). Grammaticalization in Pidgins and Creoles. The Handbook of Pidgin and Creole Studies, ed. by Silvia Kouwenberg & John Victor Singler. Malden, MA: Wiley-Blackwell, pp 385-410. Bruyn, Adrienne. (2009) Grammaticalization in creoles: Ordinary and not-so-ordinary cases. Studies in Language 33:312-337 (Special issue on 'Grammaticalization, reanalysis, and relexification in contact- induced language change'). Bureau of Ghana Languages (1977) Nunyamɔ 6. Accra. Chatelain, Heli (1888) Grammatica elementar do Kimbundu. Genebra: Charles Schuchardt. Creissels, Denis (2006) Encoding the distinction between location, source, and destination: A typological study. In Hickmann, Maya and Stéphane Robert (eds) Space in languages: Linguistic systems and cognitive categories. Amsterdam: John Benjamins, 19–28. Dereau, Léon (1955) Cours de Kikongo. Namur: Wesmael-Charlier. Essegbey, James (2005) The basic locative construction in Gbe languages and Surinamese creoles. Journal of Pidgin and Creole Linguistics 20:2, 228-267. Essegbey, James and Adrienne Bruyn (2002) Moving into and out of Sranan: Multiple effects of contact. Paper presented at International Conference on Adpositions of Movement, Catholic University of Leuven, Leuven, Belgium. Faraclas, Nicholas G. (1987) Creolization and the tense-aspect-modality system of Nigerian Pidgin. Journal of African Languages and Linguistics 3:2, 77-97. Fiaga, Kwasi (1976) Eʋegbe Ŋutinunya. Lomé: Institut de la Recherche Scientifique. Hart, René (1996) Sranantongo: Leer- en werkboek Surinaamse taal en cultuur. Arnhem: Uitgeverij Angerenstein. Haugen, Einar (1950) The analysis of linguistic borrowing. Language 26, 210-31. Heath, Jeffrey (1984) Language contact and language change. Annual Review of Anthropology 13: 367-84. Heine, Bernd and Tania Kuteva (2003) On contact-induced grammaticalization. Studies in Language 27, 529-72.
Heine, Bernd and Tania Kuteva (2005) Language contact and grammatical change. Cambridge: Cambridge University Press. Heine, Bernd and Tania Kuteva (2010) Contact and grammaticalization. In Hickey, Raymond (ed) The handbook of language contact. Oxford: Wiley-Blackwell. Heine, Bernd, Ulrike Claudi and Friederike Hünnemeyer (1991) Grammaticalization. A conceptual framework. Chicago: University of Chicago Press. Höftmann, Hildegard (1993) Grammatik des Fɔn. Leipzig: Langenscheidt. Höftmann, Hildegard & Michel Ahohounkpanzon (2003) Dictionnaire fon-français: avec une esquisse grammaticale. Köln: Rüdiger Köppe Verlag. Huttar, G.L. (1981) Some Kwa-like features of Djuka syntax. Studies in African Linguistics 12: 291-323. Huttar, George (1986) KiKongo, Saramaccan, and Ndyuka. In Elson, B. Language in global perspective: Papers in honour of the 50th anniversary of the Summer Institute of Linguistics 1935–1985. Dallas: Summer Institute of Linguistics , 563–586. INRAP (1983) Éléments de grammaire kituba. Brazzaville: Nathan Afrique. Kari, Ethelbert E. (1997) Degema. München: Lincom Europa. Lefebvre, Claire (1993) The role of relexification and syntactic reanalysis in Haitian Creole: Methodological aspects of a research program. In Mufwene, Salikoko (ed) Africanisms in Afro-American language varieties. Athens: University of Georgia Press. Lefebvre, Claire (1998) Creole genesis and the acquisition of grammar: The case of Haitian creole. Cambridge. Levinson, Stephen C (1992) Primer for the field investigation of spatial description and conception. Pragmatics 2(1), 5–47. Lumsden, John (1999) The role of relexification in creole genesis. Journal of Pidgin and Creole Languages 14, 225-258. Matras, Yaron (2009) Language contact. Cambridge: Cambridge University Press. Migge, Bettina (1998a) Substrate influence in the formation of the Surinamese Plantation Creoles. PhD dissertation, Ohio State University. Migge, Bettina (1998b) Substrate influence in creole formation: The origin of give-type serial verb constructions in the Surinamese Plantation Creole. Journal of Pidgin and Creole Languages 13, 215–265. Migge, Bettina (2000) The origin of property items in the Surinamese Plantation Creole. In McWhorter, John H. Language Change and Language Contact in Pidgins and Creoles. Amsterdam Benjamins, 201–234. Migge, Bettina (2003) Creole formation as language contact: The case of the Surinam Creoles. Amsterdam: Benjamins. Mithun, Marianne (1988) The grammaticalization of coordination. In Haiman, John and Sandra A. Thompson
(eds) Clause combining in grammar and discourse. Amsterdam: Benjamins, 331-360. Moiso, Bokula (1983) Le lingala au Zaire: défense et analyse grammatical. Kisangani: Editions du Base. Morley, Eric A. (2008) A grammar of Ajagbe. Manuscript. Mufwene, Salikoko S. (1990) Transfer and the substrate hypothesis in creolistics. Studies in Second Language Acquisition 12, 1-23. Muysken, Pieter (1981) Halfway between Quechua and Spanish: the case for relexification. In Albert Valdman and Arnold Highfield (eds) Historicity and variation in creole studies Ann Arbor: Karoma, 52-78. Muysken, Pieter (1987) Prepositions and postpositions in Saramaccan. In Alleyne, Mervyn C. Studies in Saramaccan Language Structure. Amsterdam: Instituut voor Algemene Taalwetenschaft. Muysken, Pieter (1997) Media Lengua. In Thomason, Sarah G. (ed) Contact languages: A wider perspective. Amsterdam: Benjamins, 365-426. Nyaku, F.K. (1982) Modzakaɖegbalɛ̃ enelia. Accra: Adwinsa Publications. Obianim, S.J. (1990) Eʋe Kɔnuwo. Accra: Sedco Publishing. Parkvall, Mikael (2000) Out of Africa: African influences in Atlantic Creoles. London: Battlebridge Publications. Plag, Ingo (1998) The syntax of some locative expressions in Sranan. Prepositions, postpositions or noun? Journal of Pidgin and Creole Languages 13: 335-353. Ross, Malcolm (1996) Contact-induced change and the comparative method. In Durie, Mark and Malcolm Ross (eds) The comparative method reviewed. Oxford: Oxford University Press, 180-217 Ross, Malcom (2001) Contact-induced change in Oceanic languages in north-west Melanesia. In Aikhenvald, A. and Robert M. W. Dixon (eds) Areal diffusion and genetice inheritance: Problems in comparative linguistics. Oxford: Oxford University Press. Schumann, Christian L. (1783) Neger-Englisches Wörter-Buch. Editio tertia. MS Paramaribo (Moravian Archives Paramaribo 12/648). 3 November 2013
Talmy, Leonard (1985) Lexicalization patterns: semantic structure in lexical forms. In Shopen, Timothy (ed) Language Typology and Syntactic Description, Vol 3: Grammatical Categories and the Lexicon. Cambridge: Cambridge University Press, 36–149. Talmy, Leonard (2000) Towards a Cognitive Semantics, Vol 2: Typology and Process in Concept Structuring. Cambridge, MA: MIT Press. Tavares, José L. (1915) Gramática da língua do Congo (kikongo). Luanda: Imprensa Nacional de Angola. Van de Velde, Mark L.O. (2008) A grammar of Eton. Berlin: Mouton de Gruyter. Van den Berg, Margot (2007) A grammar of Early Sranan. PhD dissertation, University of Amsterdam. Van Everbroeck (1985) Maloba ma lokóta: lingala. Kinshasa. Voorhoeve, Jan (1973) Historical and linguistic evidence in favor of the relexification theory in the formation of creoles. Language in Society 2, 133-45. Welmers, William E. (1973) African language structures. Berkeley: University of California Press. Wilner, John (ed) (2007) Wortubuku fu Sranan Tongo, 5th edition. Paramaribo: Summer Institute of Linguistics. Winford, Donald (2000) Tense and aspect in Sranan and the Creole Prototype. In McWhorter, John H. Language Change and Language Contact in Pidgins and Creoles. Amsterdam: Benjamins, 383–442.